100 likes | 444 Views
Semantic Memory. -concepts and facts. -several similar models describe the organization of semantic memory. 1) Collins & Quillian’s Hierarchical Network Model. -nodes to represent individual items, ideas, organized hierarchically. -nodes are connected by propositions (has, isa , can).
E N D
Semantic Memory -concepts and facts -several similar models describe the organization of semantic memory 1) Collins & Quillian’s Hierarchical Network Model -nodes to represent individual items, ideas, organized hierarchically -nodes are connected by propositions (has, isa, can) -where information ends up depends on cognitive economy Evidence: takes longer to respond the further the item is from a property or category: Categories: “A canary is a bird” versus “A canary is an animal” “A canary can sing” versus can fly versus has skin
Semantic Memory -problems: 1) “no” responses should take longer than “yes”, but they don’t 2) Typicality effects: “A turkey is a bird” takes significantly longer than robin 3) Mounting evidence against the categorical hierarchy -e.g. Animals—mammals—pig should be the hierarchy from top to bottom -takes longer to confirm pig is a mammal relative to pig being animal (might just be typicality again, though) -e.g.2: an animal can move, a fish can move, a shark can move “feeling”=introspection
Semantic Memory 2) Collins & Loftus’ “Spreading Activation” model -no hierarchy -’distance’ function to account for typicality -nodes activate nearby connected ones -if there is overlap in activation, you say “yes”, if not, you say “no” -organization based on experience, typicality, learning -difficult to test
Semantic Memory 3) Smith et al’s “Feature Comparison” model -no network -information for items is kept in the form of a “feature list” -comparison is made based on overlap between two lists; greater overlap=faster RT -two kinds of features: defining and characteristic -also two kinds of comparison: Stage 1 is a “global comparison” of features to get an overlap score -if very high or low, response occurs (rapid RT) -if intermediate, a Stage 2 comparison of just defining features takes place -can handle typicality: -a robin is a bird engages Stage 1, whereas a turkey is a bird necessitates Stage 2 -can handle rapid “no” responses: “a robin is a bulldozer”= almost no overlap, hence rapid response
Semantic Memory 3) Smith et al’s “Feature Comparison” model -no network -information for items is kept in the form of a “feature list” -comparison is made based on overlap between two lists; greater overlap=faster RT -two kinds of features: defining and characteristic -also two kinds of comparison: Stage 1 is a “global comparison” of features to get an overlap score -if very high or low, response occurs (rapid RT) -if intermediate, a Stage 2 comparison of just defining features takes place -can handle typicality: -a robin is a bird engages Stage 1, whereas a turkey is a bird necessitates Stage 2 -can handle rapid “no” responses: “a robin is a bulldozer”= almost no overlap, hence rapid response Problems: 1) Lack of economy 2) Lack of explanation concerning what features are defining versus characteristic Problem with ALL the theories so far: Assymetric similarity